TRANSPORTATION RESEARCH RECORD 1291 -11

Existing Methods for the Structural Design of Aggregate Road Surfaces on Forest Roads

MARGOT T. Y. YAPP, JOHN STEWARD, AND WILLIAM G. WHITCOMB

The Forest Service current!. use eve ml aggregate surfacing design methods. urrent design methods range from the "best-e. timate·· method ro technique · that were developed or adapted by the different forests or region ·. These technique have variou. defi­ciencies; the best-estimate method has been criticized bv tbe Gen­eral Accounting Office . Also. several of the ·e meihods do not provide the technical capability to analyze unusual design situa­tions. the effects of changes in the use of Forest Service roads. or the ability to incorporate technological advances. Such prob­lems as different levels of failure criteria. seasonal haul traffic. variable tire pressures, and others have al o not been met by the USDA Forest Service lllterlm Guide for Tliickne · Design of Flexible Paveme111 S1ruc111res. To meet the e need . the Forest Service recommended that a project be initiated 10 develop a second . urfacing guide and computer program for the structural design of aggr gate-surfaced and earth roads. utilizing exi. ting technology as much as pos ible . This paper focuses on a descrip­tion and evaluation of nine design methods that were deemed most suitable for use in Fore I ervice project ., The design meth­ods evaluated include all of the known method currently being used within the Fore t • ervice. The other major organization perfor~1ing aggregate- urraced or earth design. the orps of Engi­neers. ts also represented . Recommendations are made pertaining to the need for lield studies to refine the design algorithms for aggregate-surfaced and earth roads.

Because of shortcomings in the current design methods for aggregate-surfaced roads(/). the U.S . Forest Service (USFS) in 1988 reviewed the current direction regarding the de ign of such road surfacings and produced a surfacing de ign eval­uation report for internal u e and di cussion (2) . One of the key recommendation resulting from that report was to develop a surfacing. design guide for aggregate- urfaced and earth road using existing technology.

A project was therefore initiated to develop a guide and companion computer program to assist in the structural design of aggregate-surfaced and earth roads. A Forest Service Tech­nical Advisory Board consisting of representatives from sev­eral Forest Service regions was appointed to provide technical guidance during the project. The initial ta ks of the project focused on the review of existing technology related to the structural design of aggregate-surfaced roads. The results of that review and evaluation are described and the information contained in the resulting report (3) is summarized. This paper is intended as a guide to the different design methods and

M. T. Y. Yapp, ARE Inc .. 1 Victor Square. Suite G. Scotts Valley. Calif. 95066. J. Steward . USDA Forest Service. Region 6, 319 S.W. Pine Street, P.O. Box3623, Portland, Oreg. 97208. W. G. Whitcomb, Department of Public Works, P.O. Box 1995. Vancouver, Wash . 98669-1995.

their interrelationships. One of the objectives is to commu­nicate to others the nature of the research in the structural design of aggregate-surfaced roads. A similar report was sub­sequently published by the Corps of Engineers ( 4) in 1989 after the completion of this review.

Several existing relationships were identified as potential candidates for this design guide; however. no clear choice for adoption was discovered and all design methods currenrlv available had some erious limitations. In fact. there was con"­siderable disappointment with the current state of exis ing technology. However, after considerable review. a relation­ship developed by the Corps of Engineers in 1978 was selected as the thickness design algorithm for inclusion in the design guide and computer program. This algorithm has been used little, if at all, in the fields; however. this is also true for most of the design algorithms reviewed. The equation was devel­oped by researchers at the orp · Waterway Experiment Station (WES U5ing previou. field data. The inrent during it development was to provide a relationship as a starting point that could then be refined through field experiment ' .

The authors would like to emphasize that the need for field studies to refine the design algorithm is still a critical item in the development of aggregate-surfacing de. ign techniques for use by the Forest Service. Focused. mall-scale field validation experiments are vital to the continued use and acceptance of any design method selected.

Nine design methods that were deemed to be most suitable for use in Forest Service project are described. The design methods include all of tl1e known method currcmly being used within the Forest Service . The other major organiwtion performing design of aggregate-surfaced or earth roads. the Corps of Engineers, is also represented. An evaluation of the design methods and recommendations for the selected design procedure are included. Other recommendations include the desirability of field studies for refinement of the design algo­rithm · for aggregate-surfaced and earth roads. One such field study is the Central Tire Inflation Project at the Waterways Experiment Station in Vicksburg, Mississippi. which is a cooperative effort between the USFS, the Corps of Engineers. and FHWA.

DESCRIPTION OF DESIGN METHODS

Nine major design methods are summarized and evaluated:

1. U.S. Army Corps of Engineers Method; 2. USFS Region 4 implementation of Corps rutting equation ;

42

3. USFS Surfacing Design and Management System (SDMS): 4. USFS Region 8 Analysis Road Materials System (ARMS): 5. AASHTO low-volume road design method (1986) : 6. USFS Chapter 50 design method ; 7. USFS Region l seasonal surfacing method ; 8. Willamette National Forest "Seasurf" design method; and 9. FHW A report.

The design methods identified above can be categorized into three logical groups . Group 1 (items 1 and 2 above) includes work performed by the U.S. Army Corps of Engi­neers and the USFS Region 4 implementation of one of the Corps' rutting equations. In addition, a draft version of their aggregate-surfacing design guide for roads and airfields was reviewed . The three basic reports identified in this draft report are by the Corps of Engineers (5-7). Remboldt of Region 4 has adapted the design equations from Technical Report S-78-8 (5) for use in design (Remboldt, unpublished data).

Group 2 (items 3, 4, and 5 above) may be termed the SDMS group. This includes the SDMS manual and computer meth­ods, the Region 8 ARMS Program, and the 1986 AASHTO low-volume road design method (8). The ARMS program is essentially a computerization of the SDMS manual equations, while the AASHTO low-volume road design method provides nomograph solutions to the SDMS manual equations.

Group 3 (items 6, 7, and 8 above) includes three closely related methods-the USFS "Chapter 50" design method, the USFS Region 1 seasonal surfacing method, and the Wil­lamette National Forest "Seasurf' design method. The sea­sonal surfacing method uses a modified Chapter 50 design to include seasonal characteristics of materials . The Willamette "Seasurf" design method is similar to the USFS Region 1 seasonal surfacing method.

The final design procedure reviewed (item 9) is a report for FHWA (9), which contains elements from each of the above three groups and incorporates three levels of design . Level C is the simplest design method and consists of the equation 1..lt:vdoped by Hammitt (6) (Group 1). Level B is a more complex level and consists of the AASHTO design pro­cedure from the 1972 Interim Guide (10) for a one-layer sys­tem (Group 2). Level A is the most complex level and consists uf lht: 111auual de~ign procedure from SDMS (Group 3) .

The remainder of this section provides an overall perspec­tive of the foregoing design methods . All the design methods for aggregate-surfaced and earth roads found in the literature are generally related to each other and typically can be traced back to two basic studies: (a) the California bearing ratio (CBR) design method itself, developed by the California Divi­sion of Highways (JJ) for flexible pavements and adapted to airfield pavement design by the Corps of Engineers, and (b) the AASHO Road Test (12). The lineage of all subsequent aggregate-surface design methods can be traced to one or both of these studies.

Figures 1 and 2 are representative schematics of the geneal­ogy, as it were , of these design procedures from the Corps' work and the AASHO Road Test. Figure 1 shows the roots of the CBR design method with the California Division of Highways. This method was derived from empirical studies of flexible pavement performance in the 1920s and 1930s. During World War II, the Corps of Engineers was faced with the task of designing airfield pavements for the war effort.

TRANSPORTATION RESEARCH R ECOR I> /l9!

They selected California's CBR design method as a starting point and undertook an extensive series of field studies to adapt it for airfields (JJ) . The work to refine design methods for low-volume aggregate-surfaced and earth facilities has continued with the work by Hammitt (6). Barber et al. (5). and the new Corps aggregate surfacing design manual ( 7).

As shown in Figure l, the Forest Service derived its mod­ified /-factor from Hammitt's work for use in Chapter 50. This was combined with AASHTO's 1972 Interim Guide to pro­duce the 1974 version of the Chapter 50 design method . The Forest Service Region I and Willamette seasonal design meth­ods are direct descendents of the Chapter 50 design method.

Two of the three levels of design presented in the FHW A report (9) are also shown in Figure 1. The Level C design uses Hammitt's equation, whereas the Level B design uses the 1972 AASHTO Interim Guide design procedure for a one-layer system.

The work by Barber et al. (5) was a continuation of the Corps' investigations into low-volume aggregate-surfaced and earth road design. As shown in Figure 2. this research led to both the development of the manual equations in the Forest Service SDMS and the algorithms used in the computer model. Remboldt from Forest Service Region 4 also used the rutting equation to derive his design method. Both the ARMS pro­cedure (from Region 8) and the AASHTO low-volume road design method were derived from the SDMS manual equa­tions . Some modifications were made in ARMS , but the AASHTO guide adopted the equation with no modifications. Finally, in the report by Alkire for FHWA (9), these equa­tions were used as the basis for his Level A design. Each of these design methods is described in more detail below .

COE Hammitt Iii 1964-1970 - unsurf. siles

r-01r.1ugc+o.12

RU1'T1NG

CALIFORNIA DESIGN METHOD FOR

HIGHWAYS Porter, 1928-1942/ri

COE MODIFICATIONS FOR AIRFIELDS 113!

·-·~ 1940's and 1950's

FHWA - Alkire 1985 (I)

Level C l' •D.1191DgC+0.12

*USFS Chapter 50 r11 Modified r-u111oa1NirD.1JQI

w/ AASHTO 1972 101

.. ~ SEIMCEASIUTY

COE-1988 m MODIFICATIONS FOR ROADS AND STREETS

TM 5-800-10

AASHO Road Test 1121 and 1

AASHTO Interim Guide I 1972 Paved Road

Design 1101

FHWA - Alkire I'/ Level B

•Region 1 •wmamette NF Seasonal design Seasonal design

% Saturation 1---------i % Saturation I: ~-1

•AW-• Loos •·112 a1 -oo ......,, FIGURE I Design method relationships originating from California for aggregate-surfaced and earth roads.

Yapp er al. .13

Region 4 - 1988 1 - 2 - 3 Worksheet - - -·

COE 1978 Barber et al- (SJ

AASHO Road Test Data · 1950's - 1960's Serviceability design

! ESWL Modifications RD f (- ) Gravel C, > C, and Unsurf C, < C, • equation by Luhr

I Al1T11NG !

SERVICEABILm"

•,._Aggregate Loss· L.oocfs Moditiad Equation or Dir8Cl Estlmat.e

SDMS Manual Equations 1984 CSR- E, RD Rut depth and Serviceability equations ,

••soMS - Computer ' 1984 - 75 observations

Rut depth and Serviceability equation

I

ARMS- Region 8 1988 Modified SDMS Inputs Geological Maps

MSHTO LVR - 1985 1. Rutting FHWA Level A

Alkire 1985 (91

Same as SDMS Manual equations J

6 PSI, 1p instead of RD

.... .. l ... 2. Serviceability

.... '•

FIGURE 2 Design method relationships originating from'the Corps of Engineers for aggregate-surfaced and earth roads.

U.S. Army Corps or Engineers Method

This aggregate thickness design procedure is based on the Corps' early work on flexible pavements. In the 1950s. WES developed a set of design curves for flexible pavements based on the CBR (14). Mathematically, the form of the equation is

where

t design thickness (in.), f = percent of pavement design thickness,

0.23 log C + 0.15 (where C = coverages), P = single or equivalent single-wheel load (lb).

(1)

A tire contact area (in. 2) = (load/tire contact pres­sure), and

CBR California bearing ratio of underlying material.

The design thickness (r) is the thickness of the asphalt concrete and base course that would be required to protect the subgrade having a certain strength (CBR) for a number of coverages ( C) of a given load (P).

In the late 1960s, WES built test sites to develop a similar equation for determining thicknesses for unsurfaced airfields, and this is documented in Hammitt's report ( 6). The test sites contained clay materials of controlled strengths (CBR) and different depths. The failure criteria used were based on per­manent deformation or rutting and elastic deformation.

The number of coverages required to cause failure was recorded. Equation 1 can be solved for f since r, P, CBR. and A are all known. Because the test sites were unsurfaced, f is

referred to as f , or the ratio of the unsurfaced thickness to the flexible pavement thickness:

' p A [ ]

- 0.5

f = I 8.1 (CBR) - :; (2)

It was then possible to plot f against the failure coverage and, through linear regression analysis, obtain the following relationship:

f' = 0.176 Iog C + 0.12 (3)

where the terms are as previously defined. Substituting/' for fin the original thickness equation (1) results in the following:

t = co.176 log c + 0.12) [s.l <~BR) - ~J'5

(4)

This expression then determines the thickness of the cover material for unsurfaced roads required to prevent subgrade failure. The strength of the cover material is then determined (14) . The tire pressure, wheel load. and number of coverages are required to determine the CBR of the cover material.

In 1978 another study was published at WES by Barber et al. (5), which led to additional design equations for aggregate­surfaced roads. Existing rutting data from previous work at WES for earth, gravel-surfaced, and flexible pavements were utilized to develop deterioration and reliability models. Once the models had been developed, computer programs were written. The models all predicted rut depth of the surface layer given the load, tire pressure, surface layer thickness,

44

and strengths of the material layers in CBR. This study also recommended field studies to validate the relationships devel­oped ; however, conversations with Barber in 1988 indicated that the field studies were not performed.

The model developed for aggregate-surfaced roads is shown below and assumes that the top layer is stronger than the bottom layer (C, > C2 ):

(5)

where

RD = rut depth (in.), Pk = equivalent single-wheel load (ESWL) (kips). 1,, = tire pressure (psi), I = thickness of top layer (in.).

R = repetitions of load or passes. C, = CBR of top layer. and C2 = CBR of bottom layer.

In 1988. the Corps aggregate-surfacing design procedure (7) was brought in line with the existing Corps flexible pavement procedure in the interests of consistency . Figure 3 is from the

0:: C'.l (.)

100 90 BO 70

60

50

40

30

20

10 9 B 7

6

5

4

3

2

1

~

~ ~

~ ~

""'

~

~ " ~ ~ ~ ~

~

~ ~ ~ " \ " " " \ \

\

~ ~ ~ " '

TRANSPORTATION RESEARCH RECORD 1291

1988 draft version of Technical Manual 5-822-30 (7) . The design equation is

I = [ 0.1275 log(Pa~se ) + 0.897 J

[ ESWL A]" 5

x 8. 1 (CBR) - ; (6)

where

Passes = no. of repetitions of 18-kip single-axle loads; I = traffic index. a value of 2.64 is used for 18-kip

dual wheel single-axle loads ; ESWL = equivalent single wheel load (lb);

CBR = California bearing ratio of underlying material; and

A = contact area of one tire (in. 2).

This equation is similar to Equation 1, with the exception that the [-factor has been modified. In conversations with Donald Ladd at WES in 1989, it was determined that the [-factor was developed from test data that had not been published. Tests had been performed on aggregate-surfaced roads. but funds for that project were subsequently cancelled. hence the Jack

~ ~ 0

~ ~· '\~ IQ. '\.!\_ ~ " "r-..

I'\. q;,

" " '"' ~

I\. "-O-,__~ \ '\ " d'\. '\.

\ " '" r-..\\\. '\\ ~ ~\

\ " r....\\ \ \. \"\ \ \['\ \ \\ .\\ '\ \\ I\\ ~\\ ~\ ~ \

\

\~\ ~ ~ I\ J\

~ ~ ~ ~ ~ 1 2 3 4 5 6 7 B 910 15 20 30 40 50

THICKNESS, IN.

FIGURE 3 Design curves for gravel-surfaced roads (7).

Yapp et al.

of published information . At this time. the failure criteria associated with this design equation are unknown.

USFS Region 4 and Corps Rutting Equation

Rutting was selected as the failure criterion for aggregate road design in Region 4 (Remboldt . unpublished data). The design equation used in this procedure comes from work by Barber et al. (5) described previously (Equation 5). To obtain the ESWL, Region 4 has derived a regression model for use as follows:

ESWL = Cl * Lg + C2 * Lg * Da (7)

where

Cl = 0.3209 (single axles) or 0.1646 (tandem axles). C2 = 0.0151 (single axles) or 0.0127 (tandem axles),

Interaction

Lg = group load (kips), and Da = depth of aggregate (in.).

This equation is not based on the AASHO Road Test data nor that of Chapter 50. It is a new relationship based on regression equations that relate ESWL with actual load con­figurations. Remboldt indicates that the calculations used elastic-layered equations developed by Ahlvin and Ulery ( 15) .

USFS Surfacing Design and Management System

The SDMS project evolved over a period of 15 years (1972-1987) and was known by several names. including Pavement Design and Management System (PDMS). Low Volume Roads (LVR), and SDMS. In 1972 the Forest Service and the Uni­versity of Texas initiated a cooperative study to develop a pavement management system for the Forest Service road

i---------r ------T-------i--------r ------1--------, Load and Environmental Construction Maintenance Structural Operational

: Traffic Variables Vaiables Variables Design Variables Constraints r---

i Variables Variables

: I I I I I i i STAUCTURAL DE 51GN SUBSYSTEM t : Trafflc I l

i ~ Model

~ I u I Strength Model Strength Model [ti : ~ (UnP<!Ved) ! (paved) .0

l ~ ~ : :

! ! I Performance Model i Performance Model ~

I I (unpaved) (paved) l I

I

! i i .

Distr~ I

i I b_ ...... j I PSI -· Index - --- !

Time I

Time

i I .

3f3~ Wear-0u1 Function

Cost Function

TllTIEI Optimization ,____ Process

Decision Criteria

FIGURE 4 Conceptual system for low-cost roads (16) .

46 TRANS PORTA TIO.\' RESEARCH RECORD 1291

TABLE l CLASSlFJCATION OF INPUT VARIABLES FOR LVR

CATEGORY OF INPlJT VARIABLES SPECIFIC INPUT VARIABLES

MISCELLANEOUS Problem ldentif"'81ion

INPUT ~Fonnall T ota1 Number o1 Maleri111 Available Length ol Analy1ia Period Width of Each u11>e Number o1 Cards with Time Dependent llllliablM Interest Raio Road Typo (Asphalt Concrete, AQgregale, Bituminouo Surleco Troatmenl) Number ol Perl~ Poriodo Flag lor caJculalion of UMr Delay Cool

PERFORMANCE Regional Factor

VARIABLES Initial SoMceabllily Index T orminal So<vloMbillly Index lioMcoal>llly Following RahabUlllllion Non-Trolfic T""'1inal Servlceabllily Rat. of Non·Tralllc SeMoaablllly Reduction Variableo lor Aggregale l..oM Equation

TIME DEPENDENT Tralllc (Load and Froquency)

VARIABLES Maintenance Coolll Aggregate Loaa

PERFORMANCE Length ol Performance Per1oda -· Perlonnance

PERIOD VARIABLES Period is Defined u lin>e -: 1) Initial Construction and Flrll Rohabllltalion. 2) Two Major R.habllllalione. 3) Initial Conatruc:llon and Subeeq-.t Con9lrucllon ~ng the~ Typo.

CONSTRAINT Maximum Conllruotion Cool

VARIABLES Maximum Total Th~ Minimum Thic-ol lndMdual Rohabllllalion Maximum Thicl<MH ol All Rehabilitation& Maximum Thlciu- ol lndMdual Rehebllllallon Minimum Thie"'- of Top Loyer Aggtegal• l..oM lrom Eroeion (Added lo Other Aggregate Lou)

REHABILITATION Olslance Oller which Tralllc ia Slowed

PARAMETERS FOR Percent ol ADT Through Rehab Zone each Hour

CALCULATING USER P<trcent llohlclM Slopped

DELAY COSTS A\18rage Delay A- Speed Apflroecl>Jng and PUiing Through Rehab ,., .. Modal lo u.. In Caloullllng UMr Delay Coolll

GRADING AND SEAL or-. ot Seal Coal p-

COAT VARIABLES Speed ol Grader and Trucko Distance cars Follow o....., Cool of Grading or Seal Coal Time Be-n Gnadlnga and Seal Coats

VEHICLE OPERATION A- Opm1llng C<.a (Slm~lnr log Trucks and

COSTS Other than Log Truckll

MATERIAL Mmrialo.cnplion

VARIABLES Cool (NI place) per cy, ... Coelllclenl Mlnlmum llnd Maximum Loyer Thiclu>esa Selvage Value Soll Support ~or SubglWM)

network . The work was conducted in three phases and three reports were produced:

ment of a working computer-based model during Phase II . The working model was known as PDMS. In Phase III, the experiences derived from a trial implementation in several offices of the Forest Service were presented. •Phase I: conceptual study (16),

•Phase II: working model (17), and •Phase III : implementation (18) .

PDMS

In Phase I, the feasibility of developing such a pavement management system was analyzed. This led to the develop-

As a result of this third phase, two additional projects were conducted at the University of Texas. The first of these dealt with the revision of the actual design procedure used in the Forest Service and the integration of PDMS in the U.S. Forest Service road design handbook (18 ,19) . The second project was also a three-phase study to develop a data base pertaining to the design and performance of aggregate-surfaced roads. Phase I of this second project analyzed the feasibility of a

Yapp et al. 47

data base for PDMS (20). and Phase II was a pilot study (21) designed to field test alternative types of equipment for mea­suring the variables involved. recommend optimum test sec­tion length and measurement frequencies and hardware and software requirements for computerization of the data base, and develop cost estimates for several scenarios to develop the data base . In Phase III, field data were collected during 1984 and 1985 and regression analyses performed (22) .

SDMS

The work to develop SDMS took place over a 5-year period (1978-1983) and has been documented best by Luhr (23). By 1978 LVR was operating successfully. However. in July 1978. the Forest Service and the University of Texas entered into a cooperative agreement to revise and improve Chapter 50 of the Forest Service Transportation Engineering Handbook. The objective was to improve the flexible pavement design guide by making needed revisions and provide additional capabilities by accomplishing the tasks shown in the text box.

LVR

The development and implementation phases of L VR are documented elsewhere (17,1 8). The intent of the L VR pro­gram was to computerize the existing Forest Service design procedures (Chapter 50) (1) within the conceptual model pres­ented in the Phase I report (16). This conceptual model is shown in Figure 4.

These modifications initially resulted in PDMS . However. during the performance of these tasks. it became apparent that Chapter 50 was inadequate to meet the objectives shown in the text box. Therefore, a new structural design algorithm was required. Subsequently , Luhr (23) reviewed the literature and concluded that no existing design procedure was available that addressed the desired capabilities. A decision was then made to develop a new procedure using a combination of theoretical and empirical methods to specifically address the needs of the Forest Service. This new design method became known as SDMS.

The structural design subsystem was the existing Chapter 50 design. The L VR program was written to computerize this design method . The program would generate pavement struc­tures that would meet the performance limitations and con­straints provided . As design structures that met these criteria were identified, they were saved and listed in order of in­creasing total cost. The input variables for L VR are shown in Table 1.

In the initial development of the SDMS equations. Luhr (23) used an elastic-layered program known as ELSYM5 to analyze the AASHTO Road Test information and concluded

Tuks to be accomplished during SDMS project (1978-1982).

(1) OWelop r8llablllty hlclors tor dllterent daaaif'icatlona of r08da. PrcMde the dealgner wllh fleXlblllly in accordance with road'• present and future ciaaalllcillons. Provide the means tor a dMlgner to ll8lllct appropriate failure criteria tor each apecllc project

(2) Improve the Rldng Model using a IOOl'e rational apprOllCh. Improve the Aggregate Leu Model in accordance with recent research experience.

(3) Mafyze the wheel load equtvatenclee tor &Ingle and dual whaela tor legally loaded and overalzed loada • they preeently appear in Chapter 50. Analyze recent Information on .aructural layer equlvatenclea. Revile the wheel load equlvalenciea and the atructural layer equlvalenclee In Chapter 50 In llCOOfdance with thaM .nmy ....

(4) Develop a procedure In Chapter 50 tor detennining the regional factor.

(5) Provide improved procedures for evaluating the amount, type, frequency, and distribution of traffic over the design Ille of the roadway surface.

(6) Provide a capabillty to consider the effects or non-eflectS of roadway drainage.

(7) Provide a deflection design alternate in Chapter 50.

(8) Analyze the findings from recent pavement and other appropriate r8S881'ch for low-volume roads. Incorporate those Bigniflcanl findings thal area compatible with the basic approaches followed in Chapter 50. Such items as vehicle operating cost models and selected findings from the Brazil study associated with the University of Texas should be carefully llCIUtinlzed to determine their applicablllty and potential tor enhancement of Chapter 50.

(9) Provide a compreheoslve, Interactive connection between Chapter 50 and the LVR computer program. lncorporale in Chapter 50 an up-to-date LVR users manual wllh all appropriate diacussion, instruction, figures, tables, and examples of input and output data In addition, make the appropriate connections and referra18 In the LVR uuni manual with the revised Chapter 50.

(10) Check all figures, tables, nomograph charts, and other supporting material in Chapter 50 for correctness, accuracy, and ease of use. Make changes as needed. Completely rewrite Chapter so. with an appropriate explanat<lfY text to reflect all of the aforementioned changes, additions, and improvements.

48

that vertical compressive strain at the top of the subgrade was found to be the most promising parameter to correlate with the Pavement Serviceability Index (PSI). The following equa­tion (eventually used as one of the ··manual" equations for SDMS) was o:.itained from linear regression techniques:

log W" = 2.15122 - 597.622(Esc;)

( 4.2

- 1.32967 (log EsG) + log 4

_2

P,) 1.5

(8)

where

W,, number of applications of axle load x. EsG = vertical compressive strain at top of subgrade. and

P, = terminal value of PSI (terminal serviceability).

For aggregate-surfaced roads. performance models for rutting and aggregate loss were also introduced. The rutting model was obtained using Corps of Engineers data (5). Luhr indi­cates that the equation for the manual design was developed using all the data points from the Corps 1978 report. He also indicates a certain lack of confidence in the published Corps regression equations. The rutting equation is

W 18R = 0.1044 * RUT~ 575 * log 111 THICK5 155

( )

J 4J4 * (I .£8· 0~ 0) I 04k

* I,~;)() (9)

where

W 18R no. of applications of 18-kip equivalent single­axle loads (ESALS).

RUT = rut depth (in.). THICK = thickness of aggregate surface (in.).

£ 1 = modulus of aggregate surface (psi), and E~ = modulus of roadbed (psi).

In 1979, during the Low-Volume Road Conference in Ames, Iowa. a decision was made to incorporate this algorithm as a manual procedure in the "new" Chapter 50 (17). This and the development work to computerize the entire design proc­ess were completed in 1983. During the ensuing 5 years, sig­nificant changes in computer technology and existing design methods (1986 AASHTO guide) prompted the Forest Service to reevaluate the situation. In 1988 the Surfacing Design Eval­uation Report (2) was written for internal use and discussion and recommended the following:

1. Adopt the revised 1986 AASHTO design guide and the companion program (DNPS86/PC) for bituminous-surfaced roads,

2. Develop a surfacing design guide for aggregate-surfaced and unsurfaced roads using existing technology, and

3. Incorporate the concept of multiple user levels into the design process (the user levels imply differing levels of com­plexity of operation and variability of design).

USFS Region 8 ARMS Method

ARMS is a road surfacing design method developed in Region 8 of the Forest Service (Scholen. unpublished data). It was a

TRANSPORTATION RESEARCH RECORD 1291

result of the need for an aggregate management system suit­able for a regionwide situation in which there were many isolated. low-cost. short road segments. The ARMS surfacing design procedure utilizes existing geologic data (such as state geological maps) as an index to soil properties for input to surface design equations. Because of oil and gas exploration over much of Region 8. detailed geologic maps are available. The maps separate rock formations into units of similar petro­graphic characteristics, and these characteristics are then cor­related to engineering properties of soils. In this way. the maps are used to establish a geotechnical data base for use in ARMS. In addition, regression equations developed from correlated laboratory data are used to supplement traditional sampling and testing techniques.

The thickness design is based on the SDMS manual equa­tions. Two criteria are used-rut depth and serviceability loss. The design equation based on rut depth is

W1sRuT = 64.51 * (tP)- 1·•

665 * [3 (TSl)- 0·5)2·575

* log (5 . 155 * ( l~~o) 3,434 * c~~~) 1. 0-lR (lQ)

where

W 18RuT no. of applications of 18-kip ESALs. tP = tire pressure (psi),

TSI = terminal serviceability index. t = thickness of aggregate surface (in.).

£ 8 = resilient modulus of aggregate surface (psi). and EsG = resilient modulus of subgrade (psi).

Equation 10 is very similar to the SDMS manual rutting equa­tion (Equation 9). It can be seen that the last three terms are the same. The rut depth term in Equation 9 was replaced with an expression that uses terminal serviceability index instead. Region 8 has developed Table 2. which relates the rut depth as a function of Traffic Service Level (TSL). The appropriate TSI and its corresponding allowable rut depth are selected from Table 2 and entered into Equation 10.

The other criterion used in this design procedure is ser­viceability loss, and this is identical to the SDMS Equation

TABLE 2 SERVICEABILITY INDEX AND RUT DEPTH

TSL PSI TSI MRD

A 4.7 2.0 2.1

B 4.2 1.5 2.4

c 3.5 0.5 4.2

D 3.5 0.25 6.0

Notea:

TSL • Trllflic Surface Levels PSI • Preunt Serviceabillty Index TSI "" Terminal Serviceabillty Index MRD • Maximum allowable rut depth

Yapp er al.

8. The ARMS program allows the user to enter resilient mod­ulus data for both aggregate and subgrade layers. if available . If these are not known. a set of default values is available for different soil types such as crushed limestone. limerock. and sand clay. Initially. moduli were calculated on the basis of CBR using the following equation. which is found in SDMS:

MR = l.800CBR" 7 (11)

where M is the resilient modulus in pounds per square inch and CBR is the California bearing ratio.

In 1986 laboratory testing was performed on typical aggre­gates used in Region 8 to verify the assumed values. In most r.ases. the values were verified. but some were modified to reflect the laboratory results. Variations in rainfall are also considered through the number of rain days a month. Traffic is assumed to be uniform throughout the month, but the pro­gram will consider rainfall effects (and therefore weaker subgrade) in its design based on the amount of traffic hauled during rainy days. The default tire pressure is 80 psi. but other values may be entered. with the recognition that the higher tire pressures drastically increase the rate of surface failure.

AASHTO Low-Volume Road Design Method

The AASHTO (1986) design guide presents graphical solu­tions based on the SDMS manual equations (Equations 8 and 9). No modifications were made to these equations apart from converting them to nomographs. Figures 5 and 6 show the two nomographs used for design.

..__

..__

~

~

-

Exomple :

0 85 : 8 mches

E as ' 30,000 psi

MA ' 4,900 psi

CiPSI ' 3.0

Solution' w,.,,, 16,000 (18-kip ESAL)

~ ._ 3 .5

'.~ I

~ 3 .0 - """~ 25 ....

~ - 2.0

- 10 ~

'\ ~

"" ~ \.~ ~ \~

All owo blt Serv i etobll i ly ~ ~ Loss, CiPSI

\~ ~ \ ~

I\'\ \

I I I I 10 50 IOO

Allowable 18- kip Equivalent

~~ 5POO .-

to,000 I-

15,000 I-

1:0,000 ._

Sinole AKle Load Applications, W111 (thousands) '91

.jl)

However. aggregate loss has been considered. If aggregak loss is significant, an estimate may he made and added tn the aggregate thickness :

Das = D8 s + (0.5 X GL) ( 12)

where

Das total thickness of aggregate layers (in.). Dss thickness of aggregate layer based on rut depth or

serviceability loss (in .). and GL estimated gravel loss over performance period (in.).

USFS Chapter 50 Design Method

The current version of Chapter 50 was developed in 197~ and last revised in May 1982 by the office in Region 6 (1). The 1982 version was intended as an interim measure until SDMS was completed. Two sources were used to develop the design procedure. The first is the Corps of Engineers. specifically the work done by Hammitt (6) on unsurfaced roads described previously. Hammitt's work involved the construction of unsurfaced test sites and the development of a modified f factor through regression analysis. The failure criterion used was a 3-in. rut depth. On the basis of the same data. the Forest Service then similarly developed a modified/. but used a 2-in. rut depth as the failure criterion instead. The modified f that was obtained is

f = 0.216 log C + 0.1705

where C is coverages.

%1

"' ..

4

. Modulus of Bast Moteriol, ~(psi)

6 8 10 12 14

Bose Loyer Thickness, 0 85 (inches)

(13)

16

FIGURE 5 Design chart for aggregate-surfaced roads considering allowable serviceability loss (8) .

so TRANSPORT AT/ON RESEARCH RECORD 1291

Allowable 18-kip Equivalent Single Axle Load Applications, w18 (thousands)

RUT

Modulus E85 (psi)

Example'

D11 'B inchu

RD' 2 5 1nch11

MR : 4,900 psi

En ' 30,000 psi

Solution: w,.RUT' 29,000

(18-kip ESAL)

Resilient Modulus of Roadbed Material, MR (psi)

I I I I

Allowable Rut

"' O> "'

Thickness of Aggregate Base Layer Considered

for Rutting Criteria, 0 95 (inches)

FIGURE 6 Design chart for aggregate-surfaced roads considering allowable rutting (8).

The other source for the design method came from the AASHTO Interim Guide (JO), originally published in 1972 and subsequently revised in 1981 (the Blue Book). The pro­cedure for determining the weighted structural number was incorporated into Chapter 50 in 1974 with minor modifications to the nomograph. Specifically, the modifications were aimed at extending the scale for the regional factor (R) to include values greater than 5. The design eqttation for this nomograph is as follows:

log W,18 = 9.36 log(SN + 1) - 0.20

G, + 1,094

0.40 + (SN + l)S. 1~

1 + log R + 0.372(S - 3.0)

where

W,18 = total applications of 18-kip ESALs, SN = structural number,

(14)

S = soil support value, R = regional factor, and

G, = log[(4.2 - P,)/(4.2 - 1.5)] where P, is the terminal serviceability index.

Figure 7 was then developed for a given aggregate layer coef­ficient (a) of 0.14. The curved lines represent the Corps equa­tions (1, 2, and 3), and the straight diagonal line is that from AASHTO (Equation 14). The curve giving the lesser aggre­gate thickness controls. If material with a lower layer coef­ficient is used, the depth must be adjusted using the following equation:

SN = a1D 1 + a2D 2 + ...

where

SN = structural number, a, = layer coefficient for the ith layer, and D = thickness of ith layer

(15)

This last step is an unusual feature of this method, particularly if the controlling thickness was originally determined using the Corps equation.

Yapp el al.

7.0

0 0 ,...: ~

' 6.0 -1---iH-Cll-t---+-l--+-+---++----+-+---t----il---'--I

Ir 0 D..

5.0+---l-+-~-t-+--l--l---+--1--+--+-+---+.-''---+---~ (/J

...J

Si 4 .0-t---l,-t----t-t---t-+---H~-+.'-.......... t----+-- --+----1

0 5 10 15 20 25 30 35 40 45

AGGREGATE THICKNESS (in.)

FIGURE 7 Design chart for aggregate-surfaced pavements (1).

USFS Region I Seasonal Surfacing Method

In 1980 Region 1 of the Forest Service developed a design procedure for aggregate roads based on seasonal haul (24). This was referred to as the Region 1 Supplement No . 10 to the Forest Service Handbook (FSH 7709.11), Chapter 50. The modifications considered seasonal variations in subgrade soil strengths in the design of aggregate-surfaced roads. The mod­ifications could be used for shutting down haul during spring thaw, restricting haul to dry or frozen subgrades, or com­pacting subgrades to exceed standard specifications. In addi­tion, a terminal serviceability index of 0 instead of 2 was used in the design charts.

Two studies were used to monitor subgrade soil conditions and the performance of aggregate-surfaced roads. The results were used to obtain a relationship between percent saturation of the subgrade and time of the year. First, a relationship was established between subgrade soil strength, percent satura­tion, and percent compaction (by laboratory testing). Next, the seasonal variation in percent saturation throughout the design period (usually one year) was estimated from field data . Finally, the design period was divided into time periods of similar percent saturation and a CBR assigned to each increment.

Once the above steps have been completed, the aggregate thickness may be estimated for the anticipated hauling season. These roads designed with this method are for use only during certain times of the year, and are not all-weather roads . They will appear to fail prematurely if used for circumstances other than those considered in the design. Also , the procedure was developed primarily to restrict haul to periods when the subgrade is strongest ; therefore, the converse (i.e., design for when subgrades are weakest , such as during spring thaw) may not necessarily be a valid design approach .

Willamette National Forest "Seasurf'' Design Method

The Willamette design method (unpublished data, 1988) was also modified from the original Chapter 50 (/) . The modifi­cations incorporated seasonal changes in subgrade soil strengths and traffic. In addition, Miner's hypothesis is used to sum up damage ratios in the design of aggregate thickness . The pro­cedure is very similar to that described for Region l, if not the same. The regional factor used, however, is different for the Willamette . In addition , Region 1 uses a terminal ser­viceability index of 0, while the Willamette uses 2 or 1.

FHWA Report

Alkire (9) recently published a report sponsored by FHWA for the design and operation of aggregate-surfaced roads. For aggregate thickness design, Alkire provides three levels of design complexity-A, B, and C-with C being the least complex and A the most complex. Each level utilizes design methods that have generally been discussed previously . Therefore, only the key elements of the design will be high­lighted.

Level C Design

The Level C design uses the rut depth model developed by Hammitt ( 6) and is the simplest design procedure. Alkire assumes that the coverages are equal to passes and that the equivalent singfe-wheel load is 9,000 lb for an 18-kip axle load. Implied in the use of this equation is the acceptance of the 3-in . rutting failure criterion. The procedure was further

52

simplified when he suggested that the effect of traffic could be removed and thickness could be calculated using

I = (750/CBR)"' (16)

Level B Design

The Level B design is the AASHTO procedure taken from the 1972 Interim Guide (10). Figure 8 shows the relationships developed between 18-kip ESALs and structural number for various soil support values. To develop this chart the following assumptions were made: initial serviceability = 4.2; terminal serviceability = 1.5; and regional factor = 1. By using this chart, traffic may be calculated for a given thickness or thick­ness can be determined from estimated traffic. The relation­ship between structural number and thickness was shown pre­viously in Equation 15. Tabular values are also provided to adjust the layer coefficient on the basis of the strength of the granular material. In addition, guidance is given for the cal­culation of soil support and R-values.

Level A Design

The Level A design consists of the manual equations (Equa­tions 8 and 9) from SDMS. Although aggregate loss is men­tioned as a factor that needs to be incorporated into the design, no guidance is given regarding methods for estimating this factor.

Finally, Table 3 is a summary of all the design methods reviewed and their respective equations.

Aggregate Loss

Aggregate loss and its prediction are important factors in aggregate surfacing design and maintenance. One equation

2.5

a: w ID :2 2.0 ::J z -l c( a: ::J 1.5 I-(.) ::J a: I-(/)

1.0

TRANSPORTAT/Of\' RESEARCH RECORD Jl<il

that was used in the 1977 version of L VR was developed by Lund (25) and is as follows:

GL = 0.162 + 0.0188(L T) + 0.0382(F/C)

- 0.00110(TTU) - 0.00213(P3/4)

where

GL = aggregate loss corrected for.settlement (ft). LT = number of load logging trucks (OOOs).

(17)

F/C = fill or cut section (fill = 1.0, side cast slope = 1.5. cut = 2.0),

TIU = total two-way traffic units (OOOs) (one non-truck = 2TTU, one logging truck round trip = 10 TIU). and

P3/4 = percent of road surfacing sample smaller than 0. 75 in. in diameter.

Later this equation was simplified in the 1979 L VR version to

GL = 0.1 + 0.01019(LT) (18)

where terms are as identified above (25 .26). A good discussion of the Lund equations is contained in the report by McCullough and Luhr (18).

Equation 18 was used in SDMS. In addition, the SDMS user's manual provided other equations that could be used to calculate the aggregate loss manually. The first equation was developed in Brazil and the second in Kenya. The Brazilian equation is

B 3.381 GUN = - 0.0045(LADT) + -R + 0.467(G) (19)

25.4

where

GUN = aggregate loss during time period considered (in.), B = number of bladings during time period consid­

ered,

/

o.5'--~~---"..::::;;__~~~~~~~.1-~~~~~~~~.....J..~~ 1000 0 10 100

18 KIP EQUIVALENT SINGLE AXLE LOADS x 10 3

FIGURE 8 Structural number for various 18-kip equivalent axle loads and soil support values (9).

Yapp el al.

TABLE 3 SUMMARY OF DESIGN EQUATIONS

Corps at Engineers Hammitt (1970) (6) Barber et al. (1978){5) Draft TM 5-822-30 (1988) (7)

USFS Region 4 SDMS (Manual Equations)

Region 8 (ARMS) Chapter 50

Region 1 Willamette NF

FHWA Level A Level B Level c

AASHTO

SI -= Serviceability Index

Design Equation

Equation 4 Equation 5 Equation 6

Equation 7 Equation 8 Equation 9 Equation 10 Equation 13 Equation 14

Chapter 50 w/modifications Same as Region 1

Design Criteria

Rut Depth Rut Depth Unknown

Rut Depth SI Rut Depth Rut Depth & SI Rut Depth SI

SI SI

Same as SDMS Manual Equations Figure 8 SI Equation 16 Rut Depth

Same as SDMS Manual Equations

LADT = average daily traffic in design lane (for one-lane road use total traffic in both directions).

ADT = average daily traffic; N = Weinert N-value (related to three climatic levels;

precipitation and evaporation are related to the geotechnical behavior of the materials);

R = average radius of curves (ft). and G = absolute value of grade ( % ) .

The Kenyan equation (27) may have more applicability in situations where logging activity is relatively minor:

AGL= f (p: so) * 4.2 + 0.092(T)

+ 0.0889(R2) + l.88(VC) (20)

where

AGL = annual aggregate loss (in.). T = annual traffic volume in both directions of vehicles

(OOOs), R = annual rainfall (in.),

VC = average percentage gradient of the road, and f = 0.037 for lateritic gravels, 0 .043 for quartzitic grav­

els, 0.028 for volcanic gravels, and 0.059 for coral gravels.

Recent research in South Africa (28) has provided another relationship for aggregate loss:

GL = D[ADT(0.047 + 0.0027 * N - 0.0005 • P26)

- 0.365 * N - 0.0014 *PF+ 0.048 * P26] (21)

where

GL = gravel loss; D = time period under consideration (days/100);

PF = plastic limit x percent passing 200-mesh sieve (P75): and

P26 = percent passing the 26.5-mm sieve.

Finally, Paterson (29) has reviewed a number of gravel loss relationships. Table 4 shows a comparison of gravel loss rates from various studies for a range of gradient and rainfall. As can be seen, using data from other studies for the Forest Service situation can be risky, particularly for Region 6. The combinations of gradient and rainfall do not begin to resemble the road network in Region 6. Gradients of 6 to 8 percent are common, together with annual rainfalls of 1500 to 2500 mm. Paterson has included a general model to include the major traffic, gradient, and rainfall effects, but material prop­erties are excluded:

GL = (30 + 180 * MMP + 72 * MMP * G)

* h * ADT * t:..t * 10- 5

where

GL = surface material loss (mm), MMP = mean monthly precipitation (m).

G = average longitudinal gradient ( % ) ,

(22)

h = proportion of heavy vehicles in traffic (fraction) , ADT = average daily traffic (vehicles/day),

t:..t = time period (days).

54 TRAf"SPORTA T!Ol\' RESEARCH RECORD 1291

TABLE 4 COMPARISON OF GRAVEL LOSS RATES FROM VARIOUS STUDIES (29,30-34)

Rate of gravel loss (mm per 100,000 vehicle units) for gradient (%)

and annual ralQl!!ll (m!!Jarl Percentage

Study 0% 3% heavy Location Units 1,000 2,000 1,000 2,000 vehicles

Brazil v 30 32 39 43 50 Ghana HV (20-100) (10-60) 40-160 (30-60)

v 30 13 41 26 50 Oregon AV (240) 100.!/

v (40) 50 Kenya A v (7) (21) (12) (60) (35)

v 10 30 17 86 50 Kenya B v 10 29 16 82

~I Assumed value.

Notes: V = per 100,000 vehicles; HV = per 100,000 heavy vehicles when light vehicles are present but not counted; AV= articulated logging vehicles, rated as AV = 3, HV = 6, V. -. Not available. ( ). Original data not adjusted for vehicle mix.

The general effects of material properties are understood to behave as follows:

1. Increasing the plasticity index reduces the loss rate (Bra­zil, Ghana),

2. Increasing the relative compaction of the surfacing reduces the loss rate significantly (25), and

3. Coral gravels have 40 percent higher loss rates than later­itic, quartzitic, and sandstone gravels, which in turn have 40 percent higher loss rates than volcanic (vermicular) gravels (Kenya).

EVALUATION OF DESIGN METHODS

The Forest Service evaluation report of June 1988 (2) pro­vided the majority of the evaluation criteria used in this paper. These criteria may be expressed as a series of questions:

• Is the design procedure valid for aggregate-surfaced and earth roads?

• Are the inputs expected to have a major role in pavement deterioration?

•Are standard traffic units (e.g., 18-kip ESALs) used? • Can tire pressures be varied? •Is the material characterization "reasonable"? • Are risk and reliability concepts considered? •Can failure criteria levels be changed? • Is seasonal haul incorporated into the design? • Has there been any field experience?

The one criterion not included in the Forest Service report list that is included above is the validity of the design method. These items are discussed further in the following paragraphs.

Validity

The validity of the design method for either aggregate-surfaced or earth roads is probably the most important attribute to be considered. By validity is meant whether the basis of the development of these design procedures is appropriate for their purpose. As an example, it could be considered inap­propriate to design an aggregate-surfaced road using algo­rithms developed from earth road data or asphalt-surfaced road data.

Input Variables

The type and quantity of input variables are critical to the success of any design procedure. Inputs that are reasonable and recognizable to the road designer, such as CBR to describe soil strength, are highly desirable. They should be in standard units of measurement and should also be readily available. Inputs such as the internal angle of friction and approach speed of a vehicle behind a grader are either difficult to obtain, esoteric, or may seem completely unrelated to the problem at hand. The consequence of such inputs is a decrease in the usefulness of the design method to the engineer or road manager.

Traffic

Traffic data used should be in standard units, such as 18-kip ESALs, or in a format that may be easily modified to ESALs, such as timber volume (board-feet). Some design methods use units that are not easily converted into ESALs, such as the Corps of Engineers Design Index (DI). The number of

Yapp et al.

repetitions versus coverages should also be noted . because they may not necessarily mean the same thing.

Tire Pressures

In recognition of the fact that a wide range of tire pressures (as wide as 50 to 150 psi) may be used in trucks. design procedures should be able to incorporate this feature. This is particularly important because high tire pressures have a dra­matic effect in the deterioration of pavements. In addition. the Forest Service is currently conducting the Central Tire Inflation (CTI) study. which is evaluating the effects of low tire pressure on forest roads . It would be useful if the results of this study could be incorporated into the selected design procedure.

Materials Characterization

It is desirable that the materials used in the road structure be characterized in units of measurement that are standard and easily quantifiable . An example would be the use of the soil support value (S) or resilient modulus for soil strength, because these are widely known and may be available . Atterberg lim­its . although also a standard unit of measurement in soil mechanics. cannot be considered a standard parameter for the determination of soil strength in the design of roads.

Risk and Reliability

Risk and reliability may be defined in many ways. However, to evaluate the different procedures, the question posed was, " Are risk and reliability explicitly incorporated into the design method?" Currently, only the SDMS computer algorithms and the Barber et al. (5) models contain this variable. Reli­ability, as defined by AASHTO (35), is the probability that designed pavement sections will withstand the actual number of ESALs that will be applied over the design period. The method used by AASHTO is to apply a reliability design factor, FR, to the design traffic and in this way modify the thickness requirements.

Failure Criteria

The failure criteria used in most of these design methods are either rut depth or serviceability loss. An additional criterion used in several other design methods is aggregate loss. Some design procedures have developed design equations that assume a given failure level such as a 2-in . rut depth, and these may not be modified. In other methods, the failure levels are var­iables and may be changed very easily. It would be desirable to have the ability to make these changes.

Seasonal Haul

The concept of seasonal haul was developed in an effort to recognize that some forest roads are only open for use during

certain times of the year. An example would be log haul during the summer months only. with the same road being closed during spring thaw. Therefore. designing such a road for the strongest time of the year instead of the weakest would greatly decrease the aggregate thickness required. Currently. on ly a few design methods explicitly consider this .

Field Experience

Field experience is simply a question of whether each design method has actually been used for design in the field . How­ever, it should be noted that some procedures such as the AASHTO (8) guides receive such wide distribution all over the United States that it is not possible to determine if they have been used in design . The answers to this question are therefore only accurate to the best of the authors' knowledge.

Rating Scheme

Table 5 is a summary of all the aggregate surfacing design procedures included in this synthesis and their evaluation . Each of the nine attributes used for the evaluation have been discussed in the preceding paragraphs. The evaluation pro­cedure was deliberately kept simple; therefore. weights were not assigned to each attribute . If a design method contained a particular attribute, it received a plus and if it did not. it received a minus. For those that fall within the grey zone between a yes and no answer, a zero is given. The pluses and minuses are then added together to arrive at a total score .

As can be seen from Table 5, two procedures have the highest score of + 2: the Corps 1978 work by Barber et al. (5) and the Region 8 ARMS program. Note that the proce­dure by Barber et al. has more pluses than the ARMS pro· gram, although it also has more minuses. The Technical Advi­sory Board's opinion was that the pluses outweighed the minuses; therefore, the equation by Barber et al. was selected over ARMS.

Conclusions

To recapitulate, the direction for this project was to select the best available design method for use on forest roads. As noted, virtually all the design procedures reviewed, including that judged best by the Technical Advisory Board, have not been field verified. Field trials will be needed to validate the selected procedures, regardless of which procedure has been selected.

Earth Roads

At one time it was anticipated that Hammitt's (6) 1970 equa­tion (Equation 4) relating thickness to coverages and wheel load would be most appropriate for the earth road design . However, the most important information sought by the designer is a prediction of either maintenance requirements or environmental damage resulting from allowing vehicles to operate directly on the native materials. Hammitt's 1970 equa-

56 TRANSPORTATION RESEARCH RECORD 1291

TABLE 5 EVALUATION OF DESIGN MET HODS

COE -1988 1 FHWA (C) I Region 4 \ R. 8 SDMS- Manual I SDMS I USFS Region 1 FHWA(B)

ATTRIBUTES TM 5-822-3 COE • 1970 COE . 1978 I ARMS AASHTO - 1986 Computer Chap. 50 W~lamette (AASHTO 1972)

171 !Hammitt) (Barberetal.) FHWA (A) (BJ (1) (10) (6) {5) .

1 a. Validity for Aggregate

Roads + . +

1 b. Validity For Earth-Roads + +

2. Inputs Make Sense + + +

3. Standard Traffic Units

4. Varying Tire Pressure + + +

5. Material Characterization + + +

6. Risk/Reliability . 7. Change failure criteria . +

8. Seasonal Haul -9. Field Experience . .

SCORE -2 -2 +2

tion simply did not provide the means to estimate these main­tenance needs or resource impacts . Consequently , the equa­tion by Barber et al. (5) (Equation 5) was selected to predict total rut depth for the earth road situation, with some mod­ifications.

Aggregate-Surfaced Roads

On the basis of the evaluation performed and after much discussion with the Technical Advisory Board, the equation by Barber et al. (5) was selected as the thickness design algo­rithm (Equation 5). It is recognized that the algorithm has some rather serious limitations. Perhaps the most serious is that the algorithm has little, if any. field experience. However, this is also true for most of the design algorithms reviewed. Equation 5 was developed at WES through a review of pre­vious field data. The intent during the development was to provide a relationship as a starting point that could be refined through field experiments and experience in use.

J\ second major reservation that surfaced as a result of the evaluation was that the design algorithm appeared to under­predict thickness requirements for low-strength subgrade sit­uations that might simulate wet weather haul. In spite of the shortcomings and reservations , this design algorithm was selected for the following reasons:

1. The design algorithm contains most of the design factors believed to be most important by the Forest Service .

2. The equation is stable with respect to the range of design inputs selected for use. The SDMS equation was unstable for some of the ranges of input values .

3. The algorithm was the most sensitive to changes in tire pressure, which was an input criterion. In addition, the thick­nesses calculated were acceptable .

4. The design algorithm provided significantly reduced thickness requirements from those estimated through Chapter 50 for similar design inputs. This was consistent with the gen­eral perception that the Chapter 50 design method for aggre­gate surfaces is conservative.

0

0

+ + + .

+

+

+2

0 . . . .

0 . + + +

+ + + + + . .

+ . 0 0 + . + . -+ + . . .

+ - + . 0 . + + .

-1 -2 -3 -1 -4

In summary, there were clear reservations regarding the selection of the design algorithm . Unfortunately . the search of the existing thickness algorithms provided no clear choice. and it is the opinion of the Technical Advisory Board that the choice made was the best for forest road situations, given the state of the existing technology.

RECOMMENDATIONS

On the basis of the literature review and the analysis of the design methods in the preceding sections. the following rec­ommendations are made:

1. For aggregate-surfacect"and earth roads, the design algo­rithm selected was the model by Barber et al. (5) for aggregate­surfaced roads (Equation 5) . The equation was slightly mod­ified for use in the case of earth roads (the thickness of the compacted subgrade was assumed to be 6 in.).

2. One of the major disadvantages of many of the design procedures studied Wal> Lhei1 lm.:k uf fit:ld validation , partic­ularly for forest road situations. Much of the test data that was used to develop many of the procedures was, as discussed earlier, from work done by the Corps in non-forest-related projects . In the development of SDMS. the Forest Service recognized this, and attempts were made to rectify this situ­ation. It was anticipated that a data base from actual forest conditions could be used to validate or develop performance models for design of aggregate-surfaced and earth roads . However, for various reasons (e.g., insufficient traffic data) this was not successful. In the case of the Corps, projects to collect test data were also planned in cooperation with the Forest Service (5). However, they have not been followed up or the projects have been cancelled.

Therefore, the need for such field studies persists, regard­less of whatever design method is selected. It is highly rec­ommended that an attempt be made to gather some data that could be used to validate the selected design method and to enable future fine-tuning. Recognizing that data collection is expensive in time , money , and effort , it is suggested that good

Yapp et al.

use be made of any existing studies with related data. such as the Central Tire Inflation project . In addition. data col ­lection has to be flexible enough to accommodate changes in test sites. Finally. an attempt should be made to collect only data that are needed . This should include traffic : loads: subgrade and cover material strengths over seasons. particularly wet seasons: and the monitoring of the selected failure criterion . The effects of moisture on subgrade strengths are particularly important.

ACKNOWLEDGMENTS

The research described in this paper was carried out by ARE Inc .. Engineering Consultants. as part of a USDA Forest Service contract for the Region 6 office in Portland. Oregon . The authors are grateful for the support of the USFS and. in particular, Ron Williamson and the other members of the Technical Advisory Board .

REFERENCES

I. interim Guide for Thickness Design of Flexible Pavement Stnrc­tures. FSH 7709.11. Chapter 50. Region 6 Supplement No . 20 . USDA Forest Service. Portland. Oreg .. Jan. 1974. revised May 1982.

2. Forest Service Surfacing Design EFaluarion Report. Unpublished Report. USDA Forest Service. June 1988.

3. Aggregate Surfacing Design Guide and Compwer Program-Syn­thesis Report. ARE Inc., Scotts Valley. Calif.. Feb. 1989.

4. Y. T. Chou . Design Criteria for Aggregate-Surfaced Roads and Airfields . Technical Report GL-89-5. U.S. Army Engineer Waterways Experiment Station, Vicksburg, Miss .. April 1989.

5. V. C. Barber. E . C. Odom. and R. W. Patrick . The Deteriora­rion and Reliabiliry of Pavements. Technical Report S-78-8. U.S. Army Engineer Waterways Experiment Station. Vicksburg. Miss .. July 1978.

6. G . M. Hammitt II. Thickness Require111e111s for Unsurfaced Roads and Airfields. Technical Report S-70-S . U.S . Army Engineer Waterways Experiment Station. Vicksburg. Miss .. Jul y 1970.

7. Design of Aggregate Surfaced Roads and Airfields . Draft Tech­nical Manual TM 5-822-30 . Department of the Army. Washing­ton, D.C .. 1988.

8. Guide for Design of Pavement Structures. American Association of State Highway and Transportation Officials, Washington. D.C.. 1986.

9. B . D. Alkire. Design and Operation of Aggregare Surfaced Roads. Final Report. FHWA, U.S . Department of Transportation. 1985.

10. AASHTO lmerim Guide for Design of Pavement Srructures. AASHTO, Washington, D .C., 1972. revised 1981.

11. 0 . J . Porte r. Devcloprnenl on the Original Method for Highway De ign . Trawactions. ASCE. Vol. 115. 1950. pp. 461 - 467.

12. Special Report 61: The AASHO Road Tesr. HRS . National Research Council. Washington, D.C .. 1962.

13. Development of BR Flexible Pavement Design Method for Airfields. A Symposium. Trm1st1c1io11s. ASCE. Vol. 115. 1950.

14. R. G. Ahlvin . Developing a Ser of CBR Design Curves. Instruc­tion Report No. 4. U.S. Army Engineer Waterways Experiment Station. Vicksburg. Miss . , Nov. 1959.

15. R. G. Ahlvin and H . H. Ulery. Tabulated Values for Determin­ing the Complete Pattern of Stresses, Strains and Deflections Beneath a Uniform Circular Load on a Homogeneou Half Space . 811/le1i11342, HRB. National Research Council, Wa hington. D.C .. 1962.

16. T. G. lcGarrnh <Ille! W. R. Hud~on . A f>1m!111e111 Dl'.\1 1111 and M111wg t·1111t111 Sysrem fur Forc.1·1 S1•rl'icc Roads- Co11C't'Jlf l111i .\1wlr . .Research Report 12. ouncil for A1.h-a11ccd Transpon :1ti1111 Stud­ies. Univc r~ il\' of Texa at Au~tin . Jul\· I 97-1 .

17 . F. L. Roberts. B. F. McCullough. H. j , Williamstrn . and W. R. Wallin . A Pm·emenr Design and Mmwge111e/I/ ',l'.Wl'lll Ji1r Fores/ Ser1•ic1• Roads-Working Model . Research R<·po rt -13. Council for Advanced Transportation Studies . Uni\'ersity of Texas at Austin . Feb . 1977.

18. B. F . McCullough and D . R . Luhr. A Pm ·e111e111 Design a11cl :llan­agemenf System for Foresr Sen•ice Road.1-l111plc111('///lt1io11 . Research Report 60. Council for Advanced Transportation Stud­ies. Universitv of Texas at Austin . Jan . 197'1.

19. B. F. McCullough and D . R. Luhr. Th e Ne11· Chaprer 50: R<'l'i­sions IO the Transpona1ion Engineering Handbook awl Ne"' f'tll'<'-111e111 Design mu/ Management Sysrem. Draft Re port Project F!:i l-1 7709. 11. enter for Transportation Research. nive rsitY ofTcxu~.

Austin. June 1982. 20 . J.E. Hernandez. B. F. McCullough. and W. R. Hudson . A Dara

Base f ur rhe U.S. Fore r en •ia f'm ·c•1111•n1 M111111g1•111rn1 ys11•111. Transportation Research Ri:pon 66. ente r or Transpvrtatiun Rcsea.rch . Unive r itv ofTcxa~. t.-tin . M:l\' 19:!1.

2 1. Pi/01 Test of Methodology w bi! U etl for D«;·c'lopi11,f( a Daw /las<' for the Fores/ Sen•ice Pm•e111e111 Design and M111W.Qt·11w111 \ 1· ·rC'lll. EM 7170-5. USDA Fore. t Service. Si:pt . 1983 .

22. Labat-Anderson. Inc. Final Reporr 011 rlr e Forest Sen·ice Road.1 Surfacing Dma Base Projecr. USDA orcst Service. Washington Office Engineering Staff Unit. July 19 6.

23. D. R. Luhr. Develo pment of an Improved Ptw.:mcnt Design and Management System (PDMS) . Ph .D . dis crt:n iun. University or Texas at Austin. May 1984.

24. Fores/ Service Handbook 7709.J }-Region 1 Supplemenr No . JO. USDA Forest Se rvice. Missoula . Mont.. April 1980.

25 . J. W. Lund . Gifford Pi11c/101 NF Srirftrcing Lv.~ · S111cly. Unpuh­li ·hed Report . USDA For t Se rvice. ri:viscd Sept . 1977 .

26. J . W. Lund . Surfacing Loss S111dy. Region 6. USDA Forest Ser­vice, 1973.

27 . Kenya Road Transporr Cos/ Study: Research 011 Road Dererio­ra1io11 . TRRL Report 673. U.K. Transport and Road Research Labora tory. CroWl'hornc . Berkshire. England . 1975 .

28. P. Paige-Green . Some S11rfa e Ro11 •lmess . L <>ss 1111d Slipperi11es.1 Charac1erisrics of U11ptwed !loads. Proc . . !st . ympo ·ium on Sur­face Characteristics, ASTM. Philadelphia. Pa .. July 1988 (draft copy) .

29. W. D. 0. Paterson. Unpaved Roads: Concepts and Models. In Road De1eriormio11 and Maimenrmce EJ.{ecrs: Models fan· f'lm111 i11.~ and Mt111agem e111. The Highway Design and Mainte nance Swn­dards Series. World .Bank : Johns Hopkin U niven.i ty Press. Ba l­timore and London . 1 9~7. Chapter 3.

30. Rese1Jrcl1 0111/ie l111errel111io11 hips Bi!r11·11e11 °cJSls of Hi'glnmy Con­srructiun. Maime11a11ce. and Uriliwrion I /l/ RJ. Finni Report. L vols. Empresa Bra ile ira de Planejamcnt< de T rnnsport e> (GEIPOT) . Ministry of Transpo rt . Brasilin . 1%2.

31. P. W. D . H . Roberts . Performance of U11sealed Roads in Gl1t111<l . Laboratory Report 1093. U.K . Transport and Road Research Laborato ry , Crowthorne . Berk hire. England. 1983.

32 . J. W. Lund. S11rf aci11g Luss S111dy . Forc~ t Service . U.S . Depart­ment of Agriculture . Region 6. 1973.

33 . J. W. Hodges. J. Roll. and T. E. Jones. Tire Kenya Road Tram­porl Cosr Study: Research 011 Road Dereriorarion . Laboratory Report 673 . U .K. Transport and Road Research Laboratory. Crowthorne, Berkshire. England . 1975.

34. T. E. Jones. Th e Kenya Maimenance S111tly on Unpaved Roads: Reseal'clr 011 De1erioratio11. Laboratory Report 1111 . U.K . Trnns­port and Road Research Laboratory, Crowthorne. Berkshire. England, 1984 .

35 . Proposed AASHTO Guide for Design of P111•e111e111 Structure.I', Volume 2 (NCHRP Project 20-7/24). American Association of State Highway and Transportation Officials, 1985.